On Tue, May 19, 2015 at 3:01 AM, Herbert Eisenbeis <[email protected]> wrote: > My system: > Ubuntu 15.04 > Linux he-VirtualBox 3.19.0-17-generic #17-Ubuntu SMP Wed May 6 16:46:12 UTC > 2015 x86_64 x86_64 x86_64 GNU/Linux > running in virtualbox 4.3.28, 4 processors and 4GB attached > > Installing Sage 6.7 o.k., all ptestlong tests passed. > > ./sage -i p_group_cohomology stopped with: > pGroupCohomology/cohomology.c:256:22: fatal error: ccobject.h: No such file > or directory > Full log enclosed.
I hit precisely the same problem too trying to upgrade sage-6.7 for SageMathCloud on Ubuntu 15.04. What is ccobject.h? (Rhetorical question.) The reference to it in that file pGroupCohomology/cohomology.c is in some auto-generated Cython code. I don't see any explicit references to ccobject anywhere else in the package: (sage-sh) salvus@compute1-us:src$ grep ccobject -r * pGroupCohomology/cohomology.c:#include "ccobject.h" pGroupCohomology/resolution.c:#include "ccobject.h" There is a thread about ccobject, which is/was evidently part of Sage: https://groups.google.com/forum/#!topic/sage-devel/z5EtUQsI980 I tried deleting the ccobject.h line whenever it pops up in some Cython-generated code in this package (since it does in newly created files not listed above), and it gets further and hits another Cython issue: cpdef int base(self) except -1: ^----------------------------------------------------------- pGroupCohomology/mtx.pyx:773:10: Signature not compatible with previous declaration I think maybe this package needs some nontrivial attention to work. Incidentally, the next package on my list to install is something called Neuron http://www.neuron.yale.edu/hg/neuron/nrn and it's also broken in sage-6.7 (Ubuntu 15.04) apparently due to Cython issues... In building Sage+many packages for SageMathCloud, I install a bunch of packages from Pip, Sage optional packages, and R packages. The R packages always just work (which is a good thing, since I have little idea how to debug problems there). The pip/sage packages never all work. The R project must have some sort of quality control for their packages. I have the impression pypi has no quality control at all (see, e.g., [1]). With Sage optional packages, I think we have no official quality control still, but I hope that changes.... [1] E.g., there is https://pypi.python.org/pypi/pysparse, which is broken for over year according to http://stackoverflow.com/questions/25459011/how-to-build-pysparse-on-ubuntu. I guess people can just dump whatever on pypi, and there's no testing at all that it does anything? > > The following optional packages had been successfully installed beforehand: > - database_stein_watkins_mini > - database_cremona_ellcurve > - database_jones_numfield (prior: export SAGE_DATA=$SAGE_ROOT/local/share) > - database_pari > - database_symbolic_data > - gap_packages > - java3d > - gnuplotpy > - openmpi > - mpi4py > - database_gap > > Same steps had been performed with Sage 6.6: No problem to install > p_group_cohomology. > > Any clue? > > Herbert > > -- > You received this message because you are subscribed to the Google Groups > "sage-devel" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/sage-devel. > For more options, visit https://groups.google.com/d/optout. -- William (http://wstein.org) -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
